Charting the SHORES of the Radio Sky - Investigating extragalactic source populations through radio surveys

Surveys of galaxies are among the most powerful tools to investigate the Universe, allowing us to trace the statistical properties of large samples and to connect them to the physical processes driving galaxy evolution. Radio observations, in particular, have historically played a significant role — from settling the steady-state versus Big Bang debate through radio number counts, to shaping our understanding of Active Galactic Nuclei (AGN) and, more recently, uncovering obscured galaxy populations that are invisible at optical wavelengths. With the forthcoming Square Kilometre Array (SKA) and its precursors, the exploration of the faint radio sky is entering a transformative era. This thesis is part of that broad effort, focusing on the role of radio surveys in disentangling the diversity of extragalactic sources, from faint, dust-obscured galaxies to powerful blazars. It combines new survey data, population studies, and Spectral Energy Distribution (SED) analyses, and demonstrates the necessity of a genuinely multi-wavelength approach — from the far-infrared to the gamma-ray — to achieve a comprehensive view of galaxy evolution. Three complementary studies form the backbone of this work. First, SHORES, our new ATCA survey at 2.1 GHz with deep follow-up at 5.5 and 9 GHz, provided a peculiar view of the sub-mJy population. The analysis of source counts, spectral shapes, and far-infrared counterparts revealed population transitions with flux density, highlighting the interplay between AGN and star-forming galaxies and the limitations of radio-only classification. Second, in the COSMOS field, I focused on radio-selected near-infrared-dark galaxies (RS-NIRdark). These highly dust-obscured, ultra-luminous systems, at a median $z\sim3$, include extreme sub-populations such as Type-0 sources and 500 $\mu$m risers, likely tracing early, intense phases of galaxy growth. Their contribution to the cosmic star-formation rate density is substantial ($\sim$40\%), underscoring the incompleteness of UV/optical surveys and the importance of FIR and (sub)mm follow-up. Finally, the FLASH blazar analysis, based on a cross-match between ASKAP and Fermi 4FGL sources, explored the radio SED properties of bright AGN and their connection with high-energy emission. The results uncovered evidence for recurrent jet activity, extended low-frequency lobes, and strong radio–gamma correlations. This demonstrates how broadband radio data, combined with X-ray and gamma-ray observations, enhances our understanding of jet physics and evolutionary cycles. Together, these works show that the faint and bright radio skies are deeply complementary: below the sub-mJy level, multi-wavelength data are essential to distinguish AGN from star-forming galaxies, while at the bright end, correlations across the spectrum reveal the duty cycle and energetics of jets. This thesis contributes to the broader community effort to establish the systematic use of multi-band diagnostics in future surveys, paving the way for next-generation facilities — SKA, ngVLA, ALMA, JWST, and ELT — to achieve a comprehensive picture of galaxy evolution across cosmic time.